On 04/23/2015 03:28 AM, Scott Gray wrote:
"I am thoroughly familiar with Git."
"Git always screws things up."
If git always screwed things up, then those other 3 projects wouldn't be
using it.
ps: I realize this was a quote from Adrian, and not Scott.
These two statements are complete contradictions. Outcomes in git only
appear strange while you're still unfamiliar with it.
I've been using the git-svn bridge to commit to OFBiz for about 4 years and
using a git repo on my current project for the last 18 months or so.
Strange occurrences stopped happening for me after a couple of months and
generally stopped once all developers either stopped using git client UIs
that used settings they didn't understand or they started using the command
line (which is exceedingly simple for basic workflows).
I also had strangeness happen earlier on, when I first started. What I
can surmise happened, after all these years, is that I used git in the
wrong way, and git did exactly what I told it to do. But then, since I
was a noob, I didn't know what I had done was wrong, only that what I
was seeing was not what I expected.
It's the same kind of thing when you go "rm -rf $HOME". Of course all
your files are now gone, but that's not the fault of rm.
The value of feature branches and pull requests over patches cannot be
overstated IMO. The ability to easily multi-task on features, review pull
requests, keep a real commit history for contributed features, to
collaborate outside of the main repo puts git miles ahead of svn for
collaborative incremental software development.
Let's not forgot that the complete project history is available offline,
that the .svn files are not scattered all over(which makes grepping for
stuff difficult, as you have to exclude those files from results), that
you can include ancient history from previous project lifespans(I have
added svn.ofbiz.org history in one of my git-svn ofbiz clones, so I can
see history going back to 2003, well before the switch to apache, and
even when Andrew created a new repo with the mostly current component
structure).
As for that last item I mentioned, if we do switch, I would *love* to
include such ancient history.
Then, how do you(not you Scott) thing I can commit 15 changes all at
once? I do all that work in a single commit. Then I save it. Then, I
use git rebase, and split the commit into smaller chunks. Woops, that's
a new feature, let's change the order of the commits, moving that one
first. Oh, my bad, I have a typo in a commit message, let's change
that. Ok, I'm happy now, time to run all tests against every single
commit(while I switch jobs and work on something else). Ok, everything
passes, git svn dcommit $HASH, flood the mailing list.
In the svn workflow, only a single patch can be committed at a time, and
you have to manually save local work, to build up the patch history.
Git actually allows me to produce more stable code, when I am splitting
up single-large-commits.
Regards
Scott
On 20 April 2015 at 22:19, Adrian Crum <adrian.c...@sandglass-software.com>
wrote:
I am thoroughly familiar with Git. I've used it on on three projects, and
that is why I don't like it.
I have a far easier time merging branches with Subversion. Git always
screws things up.
I don't need to be convinced of anything. I have my experience and my
opinion. But still, I'm not opposed to switching to Git.
Adrian Crum
Sandglass Software
www.sandglass-software.com
On 4/20/2015 11:08 AM, Taher Alkhateeb wrote:
One of the most difficult and challenging issue with branches is _merging_
them. Git is a tool that is far more advanced in its feature set in that
area.
It seems some of the opinions expressed against git are due to
unfamiliarity. The only way to be convinced is to try it on an advanced
level as i don't think an email thread would be enough for convincing
anyone of the merits.
My 2 cents
Taher Alkhateeb
On Apr 20, 2015 12:54 PM, "Pierre Smits" <pierre.sm...@gmail.com> wrote:
If we only want GIT for multiple local development branches, then we are
doing for the wrong reasons. SVN doesn't hinder you in doing that today.
Best regards,
Pierre Smits
*ORRTIZ.COM <http://www.orrtiz.com>*
Services & Solutions for Cloud-
Based Manufacturing, Professional
Services and Retail & Trade
http://www.orrtiz.com
On Mon, Apr 20, 2015 at 11:12 AM, Jacques Le Roux <
jacques.le.r...@les7arts.com> wrote:
Like Adrian and mostly for the same reasons, I don't believe we need
Git.
But there is one other major reason which has already been discussed in
the other common ASF MLs. As Taher exulted, it's possible to create
local
branches. So people are able to do a lot of work alone without
exchanging
before committing or submitting. It will certainly not help to have this
possibility. Remember our recent discussion on the lack or core commits
reviews. With Git you end with commits bursts or big patches and it's
then
hard to review and too late to share ideas.
So unlike Adrian, I'm even strongly against it. I will not hesitate to
use
a -1 if necessary!
Jacques
Le 20/04/2015 09:53, Adrian Crum a écrit :
I don't agree that "all major contributors are using git."
Personally, I find Git to be an overly complicated solution to a simple
problem. It frequently does bizarre things that no one understands, and
you
are left with things being mysteriously reverted for unknown reasons.
This isn't a -1 vote though. I'm just making it clear that I will be
dragged kicking and screaming into using it.
Adrian Crum
Sandglass Software
www.sandglass-software.com
On 4/20/2015 5:38 AM, Hans Bakker wrote:
As discussed at apachecon in Austin, i propose to switch from svn to
git
for the ofbiz repository. The main reason being that all major
contributors are using git and contributions are cumbersome, further,
git allows for better branching and merging.
Regards,
Hans